Re: [R] almost lower triangular matrices

From: Douglas Bates <bates_at_stat.wisc.edu>
Date: Wed 02 Mar 2005 - 07:24:45 EST

Michael Anyadike-Danes wrote:
> I have output from a program which produces a distance matrix I want to
> read into a clustering program in R.
>
>
>
> The output is a .txt file and is 'almost' lower triangular in the sense
> that it is just the triangle below the diagonal.
>
>
>
> So for example a 4-by-4 distance matrix appears as,
>
>
>
> 1
>
> 2 3
>
> 4 5 6
>
>
>
> i.e. it looks like a lower triangular of a 3-by3.
>
>
>
> I thought I might be able to use "diag" to add zeros but apparently not.
>
>
>
> It's a problem because my matrix is actually 1989-by-1989 not 4-by-4
>
>
>
> I would not be at all surprised if the solution is obvious but I cannot
> quite see how to read this into R.

You can use scan to get the entries from the file in row order. Then create the matrix to hold the result and overwrite the elements in the upper triangle with scanned vector. (R stores matrices in column-major order so the row-major order from your file corresponds to the upper triangle, not the lower triangle). I'll leave it to you to work out the   symmetrization operation.

 > file.show("/tmp/tri.dat")
1
2 3
4 5 6
7 8 9 10

 > mm <- array(0, c(4,4))
 > mm[upper.tri(mm, diag = TRUE)] <- scan("/tmp/tri.dat") Read 10 items
 > mm

      [,1] [,2] [,3] [,4]

[1,]    1    2    4    7
[2,]    0    3    5    8
[3,]    0    0    6    9
[4,]    0    0    0   10
 > (res <- mm + t(mm))
      [,1] [,2] [,3] [,4]
[1,]    2    2    4    7
[2,]    2    6    5    8

[3,] 4 5 12 9
[4,] 7 8 9 20
 > diag(res) <- diag(res)/2
 > res

      [,1] [,2] [,3] [,4]

[1,]    1    2    4    7
[2,]    2    3    5    8
[3,]    4    5    6    9
[4,]    7    8    9   10

______________________________________________
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html Received on Wed Mar 02 08:48:54 2005

This archive was generated by hypermail 2.1.8 : Fri 03 Mar 2006 - 03:30:39 EST